ALPO stands for "ALPO is Like Paxos, except (hopefully) more Obvious". It is an experiment to try defining a protocol for managing a replicated state machine using distributed consensus, in a way that is easier to understand and more complete than Paxos.

Problems with Paxos

Paxos has existed for more than 20 years, is generally believed to be correct, and has been implemented numerous times. Thus it might seem silly to think about alternative algorithms. However, Paxos suffers from the following problems:

Thus, I set out to devise a Paxos-like protocol that I could understand and convince myself to be correct. If that worked, the next step is to see if I could describe the protocol in a way that others could easily understand.

Goals

Properties

In implementing the above goals, ALPO restricts the behavior of the system to have the following properties

Leader election

The first part of the ALPO protocol manages the election of leaders so that (a) there is at most one leader that can commit new log entries at any given time and (b) if the current leader crashes, a new leader will be elected to take its place. Note: the description below is slightly incomplete, since it does not describe the interaction between leader election and log management. The protocol is extended later in this document once the issue of log consistency has been introduced.

This protocol is safe because a candidate will never declare itself leader for a term unless it has received votes from a majority of the servers. Thus it is impossible for more than one server to be elected in a given term.  Furthermore, once a new leader has been elected, the old leader will be unable to create committed log entries (due to the term management protocol described below).

The deferral mechanism ensures that the protocol will converge rapidly even if there are initially many candidates. Once a candidate has deferred, it will not become a candidate again until the timeout period has elapsed, and the timeout period is large enough to get through several election cycles. In addition, the timeout period is restarted whenever a follower receives a message from a candidate, which further reduces the likelihood that a candidate will reenter an election once it has deferred. Thus it is unlikely to take more than two election cycles to select a new leader.

Log management

This section describes how the log is managed during a particular term, and how log consistency is preserved  when leadership changes.

Replicating new entries. Clients append to the log by making requests through the leader. The leader adds the new entry to its log, then sends an appendEntry request containing that log entry to each of the other servers. Each server appends the entry to its log and also writes the data to secondary storage; once this is done, the server is said to have accepted the log entry. Once a majority of the cluster has accepted the new log entry, the leader can respond to the client.  At this point the entry is called committed because its durability is assured; the only event that could cause it to be lost is simultaneous catastrophic failures of more than half the servers in the cluster (i.e. they lose their secondary storage permanently).

Follower crashes. If a follower crashes then it will not be able to accept new data from the leader. The leader need not wait for the crashed server to restart before responding to client requests: as long as a majority of the cluster is responsive, the cluster can continue operation. When a server restarts after a crash, it enters follower mode (it does not attempt to contact the leader). If the leader does not receive an acceptance from a server when it sends a new log entry, it continues trying at regular intervals; eventually the server will restart, at which point the leader will "catch it up" with the log entries it has not yet accepted. This mechanism ensures that all servers will eventually mirror all log entries.

Leader crashes. Leader failures are more interesting. At the time of a leader failure, there may be one or more log entries that have been accepted by a few nodes but not yet committed (i.e., the leader has not yet responded to the requesting client). There may also be any number of log entries that have been committed by the cluster, but are not yet fully replicated on all servers. For the partially accepted entries, the new leader must guarantee that these entries are either fully replicated, or completely expunged from all logs. Any entry that is expunged must never be returned to a client in a read operation: the system must behave as if the client never made its original request. For entries that have been committed but not fully replicated, the new leader must make sure that these entries are eventually fully replicated.

ALPO uses a two-step approach to handling leader crashes.  First, it makes sure that the new leader's log is complete (it includes all of the committed entries).  It does this by choosing the new leader from among those servers whose log is complete. ALPO extends the notion of rank during leader election to include log length. When candidates request votes they include in the request the log id of the most recent entry they have accepted. One candidate automatically outranks another if its "last accepted id" is higher than that of the other candidate. If they both have the same "last accepted id", then rank is determined by server id. Furthermore, a server will automatically reject a request for its vote if its "last accepted id" is higher than that of the requesting candidate, even if its vote is still available; when a candidate receives this form of rejection it drops out of the election and returns to follower state. These rules guarantee that the new leader's log is complete: the new leader must have contacted a majority of servers in the cluster, and its log must be at least as complete as any other log in that majority, so all committed entries must be in the leader's log.

The second step in handling leader crashes is ensuring log consistency of the followers. When a new leader takes over, its log is up-to-date, but logs on other servers may be inconsistent in either or both of two ways:

In order to guarantee consistency of the log on all nodes, a follower only accepts a new log entry from the leader if it can be sure that all of its entries preceding the new one match exactly the corresponding entries in the leader. If accepting the new entry would violate this rule, then the follower rejects the appendEntry request and returns a log id to the leader; if the leader sends the entry indicated by the follower, followed by all entries after that one, the follower's log will eventually become identical to the leader's log. It's possible that in some cases the follower may reject several appendEntry requests in a row, but eventually this process will identify the log prefix where the leader and follower are identical, and the leader will then send the entries after this prefix.

In order to implement this rule, each log entry stores the term in which it was accepted by the leader. Each appendEntry request includes the term of the previous entry in its log (in addition to the term of the new entry and its log id). When a follower receives an appendEntry request, it first removes from its log any entries with ids greater than or equal to the id of the new entry from the leader, since these are clearly extraneous. Next, the follower makes the following checks:

Managing terms

This section contains more detailed information on managing terms. Terms are used to distinguish votes from different election cycles, and also to detect out-of-date servers and extraneous log entries.  In general, if a candidate or leader finds itself out of date it immediately passivates, under the assumption that another server is leader or will become leader soon.

Log cleaning. One final issue related to log management is log cleaning. ALPO allows each server to perform cleaning (or any other form of log truncation) on its log independently of the other servers. However, there is one restriction on log cleaning: a server must not delete a log entry until it has been fully replicated. Otherwise the server could become leader and need that entry to update a lagging follower. To ensure this property, the leader keeps track of the highest log id that has been fully replicated and includes this value in any requests that make to other servers. The other servers use this information to restrict cleaning; in most cases the fully-replicated-id will be at or near the head of the log, so this will not impose much of a restriction on cleaning.

Clients: exactly-once semantics

In ALPO, clients must send any requests that result in log modifications to the leader. If such a request arrives at a follower (for example, because it used to be leader but has been deposed) the follower rejects the request; in most cases it will be able to tell the client who is currently the leader. Clients can send read requests to any server in the cluster. However, followers may not be able to return the most recent log entries, for two reasons. First, the server might not have accepted the most recent log entries yet. Second, only committed log entries can be returned to clients, and a follower may not know whether its most recent log entries are committed. One way to handle this is for the leader to include the highest committed id in each request to other servers; in most cases, the append for entry N would indicate that entry N-1 is now committed, so the follower would would lag at most one log entry in comparison to the leader. Thus, if clients can tolerate a small amount of lag they can issue log reads to any server; if they want to be assured of getting all the most recent data, then they must send requests to the leader.

ALPO can provide exactly-once semantics to clients, meaning that if the leader fails while processing a log append request from a client, the client library package can automatically retry the request once a new leader has been elected, and the new entry will be committed to the log exactly once (regardless of whether the original request completed before the leader crashed). In order to implement exactly-once semantics, clients must provide a unique serial number in each request; this serial number, along with the client identifier, must be included in every log entry so that it is seen by every server. Using this information, each server can keep track of the most recent serial number that has been successfully completed for each client. When a client retries a request because of leader failure, the new leader can use this information to skip the request if it has already been successfully completed.

Log cleaning

One final issue related to log management is log cleaning. ALPO allows each server to perform cleaning (or any other form of log truncation) on its log independently of the other servers. However, there is one restriction on log cleaning: a server must not delete a log entry until it has been fully replicated. Otherwise the server could become leader and need that entry to update a lagging follower. To ensure this property, the leader keeps track of the highest log id that has been fully replicated and includes this value in any requests that make to other servers. The other servers use this information to restrict cleaning; in most cases the fully-replicated-id will be at or near the head of the log, so this will not impose much of a restriction on cleaning.

Requests

There are only three requests sent between servers in ALPO:

appendEntry(entry, prevTerm): add entry to the local log, assuming that its id is one greater than that of the last entry currently in the log, and that the term of the last entry in the log matches prevTerm.

heartbeat(leaderId, term, lastLogId, lastLogTerm): used to inform followers that there is a leader and that the leader is alive and well. LeaderId is the id of the leader. The lastlogId and lastLogTerm arguments specify the id and term from the leader's last log entry; if these don't match the last log entry in the recipient, then it returns the id and term from its last log entry so the leader can replay missing entries.

requestVote(serverId, term, lastLogId, lastLogTerm): used by a candidate to request another server's vote for term.  ServerId is the id of the requesting candidate, and lastLogId and lastLogTerm specify the id and term from the requestor's last log entry; if either of the recipient's corresponding items is larger then it means the requestor's log is not complete enough for it to become the new leader, so the request is rejected.

State for each server

Additional state kept by leader

Contents of a log entry

Important parameters